Let’s read Julia documentation in your preferred language
SatoshiTerasaki@AtelierArith
Link to this slide
https://atelierarith.github.io/AtelierArith_JuliaCon2025_talk/slide/slide.html#/title-slide
Overview
- Introduction to DocstringTranslation.jl
- Multilingual translation of Julia documentation
- Running demonstration
Background and Challenges
- When you want to something to learn about Julia, we need to read Julia’s official documentation in English
if you in Set(NativeEnglishSpeakers)
println("No problem :D")
elseif you ∈ Set(GoodAtReadingEnglish)
println("It's O.K.")
else
lang = native_language(you) # e.g., Japanese
@assert lang ≠ "English"
println("Meh.")
println("It takes a lot of time to understand correctly.")
println("I want to read documentation in $(lang)")
endBackground and Challenges
- All up-to-date information is provided only in English
- It’s O.K. English is a common language
- For non-English-speaking developers, reading and participating in discussions in English represents significant psychological and time costs
- This is a yet another two-language problem
- Manual translation processes require significant time
Challenges (in Japan)
- The number of people who use English daily is limited. Many technical documents are expected to be readable in Japanese.
- In Japan especially in industry, C++/Python is widely used in industry. There is lots of Japanese documentation (including blog post) for C++/Python.
- “I don’t want to use Julia because it is a minor language” C++ user says
- “Julia is difficult” C++ user says
- “Few libraries and documentation available for Julia” Python user says
- “Production systems must strictly prohibit the use of the Julia language. From now on, if you use it, be prepared for possible termination.” CEO at my workplace (This is NOT a joke.)
Challenges (in Japan)
- Starting August 2025, I’ll be unemployed. This Is Not a Drill.
- To grow the Julia community in Japan, I think we need documentation written in Japanese. That’s why I created DocstringTranslation.jl package.
Solution
- Developing an automated translation system utilizing machine translation
What is DocstringTranslation.jl?
A package for translating Julia docstrings using OpenAI API
@doc exp (Original)
julia> @doc exp
exp(x)
Compute the natural base exponential of x, in other words ℯ^x.
See also exp2, exp10, and cis.
Examples
≡≡≡≡≡≡≡≡
julia> exp(1.0)
2.718281828459045
julia> exp(im * pi) ≈ cis(pi)
true
exp(A::AbstractMatrix)
Compute the matrix exponential of A, defined as:
e^A = \sum_{n=0}^{\infty} \frac{A^n}{n!}.
For symmetric or Hermitian matrices A, an eigendecomposition (eigen) is used; otherwise, the scaling and squaring algorithm
(see [^H05]) is employed.
│ [^H05]
│
│ Nicholas J. Higham, "The squaring and scaling method for the matrix exponential revisited", SIAM Journal
│ on Matrix Analysis and Applications, 26(4), 2005, 1179-1193. doi:10.1137/090768539
│ (https://doi.org/10.1137/090768539)
Examples
≡≡≡≡≡≡≡≡
julia> A = Matrix(1.0I, 2, 2)
2×2 Matrix{Float64}:
1.0 0.0
0.0 1.0
julia> exp(A)
2×2 Matrix{Float64}:
2.71828 0.0
0.0 2.71828@doc exp (Translated into Japanese)
julia> ENV["OPENAI_API_KEY"] = "sk-<blah>"
julia> using DocstringTranslation
julia> @switchlang! :ja
julia> @doc exp
exp(x)
xの自然基底指数を計算します。言い換えれば、ℯ^xです。
他にexp2、exp10、およびcisも参照してください。
例
≡≡
julia> exp(1.0)
2.718281828459045
julia> exp(im * pi) ≈ cis(pi)
true
exp(A::AbstractMatrix)
行列 A の行列指数関数を計算します。定義は次の通りです。
e^A = \sum_{n=0}^{\infty} \frac{A^n}{n!}.
対称行列またはエルミート行列 A
の場合は、固有分解（eigen）が使用され、それ以外の場合はスケーリングと平方化アルゴリズム（[^\H05]を参照）が選択されます。
│ [^H05]
│
│ Nicholas J. Higham, "The squaring and scaling
│ method for the matrix exponential revisited",
│ SIAM Journal on Matrix Analysis and
│ Applications, 26(4), 2005, 1179-1193.
│ doi:10.1137/090768539
│ (https://doi.org/10.1137/090768539)
例
≡≡
julia> A = Matrix(1.0I, 2, 2)
2×2 Matrix{Float64}:
1.0 0.0
0.0 1.0
julia> exp(A)
2×2 Matrix{Float64}:
2.71828 0.0
0.0 2.71828